Customer Feedback Synthesizer

You are reading customer reviews one tab at a time. G2 on Monday, Trustpilot on Wednesday, App Store exports that never get opened. You paste quotes into a doc, tag them manually, and by the time you have enough to say something, the sprint planning meeting already happened. The feedback exists. The signal does not. The Customer Feedback Synthesizer is a structured workflow skill that ingests multi-source review data and produces a ready-to-present synthesis: sentiment scores by source, clustered themes ranked by frequency, NPS trend lines, and a feature request stack ranked by mention volume. Built from 80+ real support and CS review cycles — every aggregation rule exists because manual tagging produced the wrong priority call. Unlike a spreadsheet template or a generic summarization prompt, this skill never collapses all negative feedback into a single "dissatisfied customers" bucket — that anti-pattern buries the difference between a billing UX complaint and a churn-driving product gap. Theme clustering runs at the sub-sentiment level so product and CS teams see distinct, actionable signals, not averaged noise. What you get: FEEDBACK_SYNTHESIZER.md — the full workflow document covering ingestion, scoring, clustering, and output formatting. THEME_CLUSTER_TEMPLATE.md — a structured tagging schema for mapping raw reviews to ranked themes. NPS_TREND_TRACKER.md — a repeatable scoring and trend-line format for monthly or quarterly reporting. Requires no external API keys — runs inside any Claude-based agent or OpenClaw instance.